Hidden assumptions

Earlier in the week, I stumbled upon this article, Unearthing deeply rooted plant 'myths' . It is about Dr. Henry Oakeley, the garden fellow for the London's Royal College of Physicians, and his view about herbal medicine.
He posits that most plants are actually trying to harm us, or at least are neutral, going against the usual view that "herbals" are more "natural" and therefore better for us.
While reading the article, I realized that I had never thought about this before. I have evaluated the evidence for specific herbal remedies (and found most of them wanting, from the scientific standpoint), but I never thought about it in a global way. I even had the hidden assumption that we have to prove that they are not useful/that they are beneficial as my baseline, instead of having a probability based assumption.
Let me explain that better.
When you have a large variation for a character(the place in the beneficial-harmful spectrum for hundreds of thousands plant species), and very tight constraints to fulfill (human biology) it is logical by probabilities that most of the variations will be either null or harmful (similar to the relationship between mutations and organisms). Therefore multiple generations of humans had to go to the trial and error process to find the ones that are useful. And these would have been very few, relatively speaking.
But then, why do we have this assumption that "herbals" are good for us? Human selection and availability bias, I would say. Whenever we find  a plant that benefits us in any way, we will select it and plant more of it around us. Therefore we see them more, and when we have to think about the relationship between plants and us, we will think about the ones that we plant ourselves, and give them predominance in our mental representation of how plants interact with us. Dr. Oakeley is able to escape this assumption, because he takes care of thousands of different varieties of plants, and not only for alimentary or health reasons. He is able to realize how infrequent are the ones that actually are useful to us.
This illustrates how even when you are trying to be rational about a topic, like I was when I  evaluated the claims of different herbal medications, you may have some non manifest assumptions that may color not only your perception, but how you interact with that aspect of the world. It makes even more important both the systematization of how you evaluate claims, but also your exposure to people that have very different backgrounds, to use them as a detection machine for your own hidden assumptions.